RGG1 Encodes a Type-A Gγ Subunit
In the rice genome, five Gγ subunits have been identified. Among them, RGG1 is relatively small and contains four exons (Fig. 1a). Phylogenetic analysis showed that the G proteins of rice, Arabidopsis, and maize were divided into three groups (Fig. 1b). Among types A, B and C, the amino acid sequences showed very little conservation, and most of the similarities were limited to a highly conserved GGL (G gamma-like) domain (Fig. S1). RGG1 belongs to a clade of type-A G proteins along with the AGG1 and AGG2 proteins of Arabidopsis. SMART analysis predicted that RGG1 contains a nuclear location signal (NLS) at the N-terminus, a GGL domain, and a CaaX isoprenylation motif at the C-terminal end, typical of all canonical type-A G proteins (Fig. S1).
We confirmed that RGG1 interacts with RGB1 using a bimolecular fluorescent complementation (BiFC) assay. The BiFC fluorescence signal was detected in the membrane, cytoplasm and nucleus, suggesting the potential function of the Gβγ dimer (Fig. 1c). To further explore the interaction of RGG1 and RGB1, several truncated RGG1 proteins were generated. As shown in Fig. 1d, the GGL domain was necessary and sufficient for interaction with RGB1. In addition, residues 55–67 of RGG1 were required for the RGG1-RGB1 interaction (Fig. 1d).
Expression Profiles and Subcellular Localization
To determine the expression pattern of RGG1, the tissue-specific expression of RGG1 was detected using transgenic plants containing an RGG1 promoter: GUS fusion. GUS staining revealed different levels of expression of RGG1 in panicles at different developmental stages. As shown in Fig. 2a, the expression of RGG1 gradually decreased with panicle development. It was also expressed in roots, with particularly strong staining in the root tips (Fig. 2b). Additionally, GUS staining showed that RGG1 was abundantly expressed in leaves, sheaths, nodes, stems, and spikelets (Fig. 2c-h). Moreover, the GUS results were in agreement with our quantitative reverse transcription-PCR (qPCR) analyses, which showed particularly high expression of RGG1 in young panicles and decreasing panicle expression as development progressed (Fig. 2i). In addition, we detected RGG1 transcripts in other tissues using qPCR, including leaves, stems, nodes, sheaths, and roots (Fig. 2i). These data suggest that RGG1 may play an important role in panicle and seed development.
To observe the subcellular localization of RGG1, both green fluorescent protein (GFP) and an RGG1-GFP fusion protein driven by the CaMV 35S promoter were transiently expressed in rice protoplasts. Similar to the GFP signal, RGG1-GFP was detected in the plasma membrane, cytoplasm and nucleus (Fig. 2j). To verify the function of the predicted NLS at the N-terminus, we also transiently expressed a truncated protein, RGG1ΔNLS-GFP, in rice protoplasts. However, the fluorescent signal of RGG1ΔNLS-GFP showed the same distribution as that of RGG1-GFP, suggesting that the putative NLS domain may not be functional (Fig. S2).
Overexpression of RGG1 Resulted in Yield Reduction in Nipponbare Rice
To elucidate the biological function of RGG1, overexpression and knockout vectors were generated and then transformed into NIP using an Agrobacterium tumefaciens-mediated method. Several successful transformed lines were obtained and confirmed by qPCR and sequencing. We chose two overexpression (OE) and two mutant lines for further analysis (Fig. 3a, b).
The relative expression levels of RGG1 in two OE lines (OE1 and OE2) were detected. Compared to that in NIP, the expression level of RGG1 was higher by eight- and twelve-fold in OE1 and OE2, respectively (Fig. 3b). As a result, the OE1 and OE2 transgenic lines showed a semi-dwarf phenotype at maturity (Fig. 3c). Further analysis showed that all the internode lengths of the OE lines were shorter than those of NIP (Fig. S3a, b). Additionally, we quantified other yield components, such as panicle length (PL), tiller number per plant (TN), GN and 1000-grain weight (TGW) (Fig. 3h, i, l, Table S1). Neither PN nor TN showed a difference between NIP and the two OE lines (Fig. 3h, i). However, the TGW values of OE1 and OE2 decreased by 19.20% and 19.44%, respectively, compared to that of NIP (Fig. 3l). Further analysis suggested that RGG1 affects grain length and width but has no influence on grain thickness (Fig. 3e-k, Table S1). In particular, the grain lengths in the OE lines were lower by 7.41% and 10.17%, respectively, than that in NIP (Fig. 3j). As expected, OE1 and OE2 also exhibited decreased grain yield per plant (Fig. 3m). Taken together, these results indicate that overexpression of RGG1 can cause semi-dwarf height and shortened grain length.
Additionally, two knockout mutants of RGG1 were generated using the CRISPR/Cas9 system in the NIP background (Fig. 3a). Sequencing results showed that both mutants, NIP-rgg1–1 and NIP-rgg1–2, had large deletions in the target site that abolished protein expression. However, the mutant plants of lines NIP-rgg1–1 and NIP-rgg1–2 did not show any obvious phenotype in traits including plant height and other yield component (Fig. 3c-m). This result may be due to the extremely low expression level of RGG1 in NIP (Fig. 3b). Whether the role of RGG1 in signal transduction is subject to genetic redundancy needs further study.
Overexpression of RGG1 in Wunyunjing 30 Results in a Similar Phenotype
To investigate whether RGG1 shows similar effects to those in the NIP mutants in the qpe9–1/dep1 mutant background, we transformed Wunyunjing 30 (WYJ30), a high-yield variety of rice that naturally lacks a functional qpe9–1/dep1, with the RGG1 overexpression vector, and we measured plant height and other yield-related traits at maturity. Both the WYJ30-OE1 and WYJ30-OE2 lines showed reduced plant height and PL compared to those of WT-WYJ30 (Fig. S4a-d, Table S2). Additionally, compared to WT-WYJ30, the grain lengths of the two OE lines were reduced by 3.29% and 3.42%, respectively (Fig. S4e). There was no significant difference in grain width between the WYJ30 and OE lines (Fig. S4f). In contrast, overexpressing RGG1 caused decreased TGW and grain yield in WYJ30 (Fig. S4g, h). These results suggest that the roles of RGG1 in regulating plant height and grain length are independent of qPE9–1/DEP1. Pyramiding different Gγ-encoding genes may be a suitable way to modulate grain size in rice.
We also used the CRISPR/Cas9 method in WYJ30 to obtain several homozygous mutants of RGG1. We identified one line, WYJ30-rgg1–1, with a 4-bp deletion and another with a 1-bp insertion, WYJ30-rgg1–2 (Fig. S5). Both mutations disrupted the GGL domain (Fig. S5). We observed no changes in plant morphology or grain size between WT-WYJ30 and these two mutants (Table S2). Taken together, knockout of RGG1 might not affect rice growth and development.
RGG1 Regulates Grain Size by Affecting Cell Division
The spikelet hull has an important impact on grain size determination. Compared with WYJ30, both OE lines had reduced grain lengths and grain widths (Fig. 4a, b). Generally, organ size is determined by cell expansion and division. To investigate the grain size differences between the WYJ30 and OE lines, histological cross-sections of the spikelet hulls were analysed (Fig. 4c-e). As shown in Fig. 4d and e, both the OE lines had significantly higher cell areas and lower cell numbers than WYJ30. Furthermore, the epidermal cells of WYJ30 and the transgenic lines were analysed using scanning electron microscopy (SEM) (Fig. 4f). No obvious difference in cell length or cell width was found between the WYJ30 and OE lines (Fig. 4g, h). However, the OE lines had fewer longitudinal cells than WYJ30 (Fig. 4i). Overall, these results suggest that overexpressing RGG1 suppressed cell division in the spikelet hull and consequently led to smaller grain size.
RGG1 is Involved in Cytokinin Biosynthesis
Due to the significant influences of RGG1 on panicle elongation and grain length that we observed, we then performed a transcriptome analysis to investigate the possible molecular pathway of RGG1 action in the young panicles of NIP, NIP-rgg1–2, and NIP-OE2. A total of 1463 differentially expressed genes (DEGs) were detected in OE2 compared with NIP; 690 of these genes were upregulated, and 773 genes were downregulated (Fig. 5a). Additionally, 754 DEGs, including 249 up- and 505 downregulated genes, were found in the young panicles of the rgg1–2 mutant. The detected DEGs were involved in diverse biological processes and metabolic pathways (Fig. S6). Analysis of the DEGs using Gene Ontology showed that the greatest enrichment was in the biological process category. Additionally, Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses revealed that the zeatin biosynthetic pathway was enriched in DEGs (Fig. 5b). In particular, many DEGs were associated with cytokinin biosynthesis (Fig. 5c). Notably, one gene, LOC_Os01g40630, encoding the cytokinin-activating enzyme LOG, which is responsible for converting inactive cytokinin to biologically active forms, was downregulated in the young panicles of NIP-OE2 (Fig. 5c, d). The expression levels of several cytokinin biosynthetic genes were confirmed using qPCR assays (Fig. S7a-c), and LOG was found to be downregulated in the young panicles of the OE lines and upregulated in rgg1 mutants. We also analysed two other cytokinin biosynthetic genes. CYP735A4, encoding the key enzyme converting isopentenyladenine (iP)-type to trans-zeatin (tZ)-type cytokinins, had decreased expression in the panicles of the OE lines, while no significant changes were observed in the mutants. OsIPT9 encodes IPP transferase for synthesizing cZ in rice (Tsai et al. 2012) and was downregulated in the young panicles of the OE lines and upregulated in the rgg1 mutants. These results suggest that RGG1 might be involved in the cytokinin regulatory pathway.
To test this hypothesis, we measured the concentrations of cytokinin in young panicles (Fig. 5e-h). The total contents of two cytokinin precursors, N6-(Δ2-isopentenyl) adenosine (iPR) and trans-zeatin-riboside (tZR), in the OE lines were similar to those in NIP (Fig. 5e, f). However, the contents of the active forms, iP and tZ, were significantly lower in the OE lines than in NIP (Fig. 5g, h). NIP-OE2 accumulated more tZR than NIP did, and this effect may be due to an inefficient conversion ability (Fig. 5f). These results suggest that overexpression of RGG1 reduced the efficiency of the conversion of cytokinin precursors to active forms, possibly as a result of lower expression of LOG or other genes in the cytokinin pathway.
RGG1 Affects Cytokinin Signalling
Heterotrimeric GTP binding proteins (G proteins) are involved in multiple signal transduction processes and intracellular responses to stimuli in plants. We also investigated whether RGG1 affects cytokinin signal transduction in rice. Shoot and root elongation assays were conducted to test the sensitivity of the overexpression and mutant lines to different concentrations of 6-benzylaminopurine (6-BA) (Fig. 6a). These experiments revealed an altered growth curve for the RGG1 overexpression lines when treated with 6-BA. At low concentrations, the shoot elongation of NIP and the two mutants was more strongly inhibited than that of OE1 and OE2 (Fig. 6b). The inhibition of root elongation by cytokinin was also compared between NIP and transgenic lines (Fig. 6c). These results showed that the two OE lines had reduced sensitivity to 6-BA with respect to its inhibitory effect on root elongation (Fig. 6c). All these results indicated that RGG1 is involved in cytokinin biosynthesis and signal transduction in rice.